Analysis of behavioral effects of drugs

Drug Development Research 20:389-409 (1990)

Analysis of Behavioral Effects of Drugs Jeffrey M. Witkin and Jonathan L. Katz

Psychobiology Laboratory, NIDA Addiction Research Center, Baltimore, Maryland

ABSTRACT

Witkin, J. M., and J.L. Katz: Analysis of behavioral effects of drugs. Drug Dev. Res. 20:389-409, 1990.

Behavioral pharmacology has largely replaced explanations of the behavioral effects of drugs in terms of hypothetical constructs with descriptions of the environmental or behavioral factors that modify drug effects and by characterization of behavioral mechanisms of drug action. The qualitative and quantitative modulation of behavioral effects of drugs by current and past behavioral and environmental variables has been amply demonstrated. However, fewer studies have been directed toward elucidation of behavioral mechanisms of drug action. The concept of behavioral mechanism implies that drugs can alter behavior in a manner functionally equivalent to the manner in which other variables alter behavior. Owing to the complexity of the determinants of behavior, a host of tactics is required for a comprehensive analysis of the behavioral effects of drugs. Detailed understanding of the various interpretations associated with the results of descriptive and mechanistic studies is essential to the development of adequate accounts of the manner in which drugs affect behavior. Studies elucidating the behavioral or environmental factors that influence the behavioral effects of drugs as well as studies of behavioral mechanisms of drug action may aid other disciplines within pharmacology and the neurosciences by ultimately pinpointing specific behavioral substrates of drug action. This information should greatly facilitate further investigations of the link between behavior and physiological processes as well as provide rational bases for drug development.

Key words: behavioral pharmacology, experimental analysis, mechanisms of action, environmental factors, rate dependency, anxiolytic drugs

Received final version December 20, 1989; accepted January 25, 1990.

Address reprint requests to Dr. J.M. Witkin, NIDA Addiction Research Center, P.O. Box 5180, Balti- more, MD 21224.

0 1990 Wiley-Liss, Inc.

390 Witkin and Katz

INTRODUCTION

For many pharmacologists, mechanisms of drug action are to be found in the unique cascade of biological events that take place upon drug administration and their triggering by precise pharmacological recognition processes. Although behavioral pharmacologists do not dispute the importance of these mechanisms, they emphasize additional variables as integral parts of any description of the determinants of the behavioral effects of drugs. The need for these additional factors stems from observations that drug effects on behavior are not immu- tably linked to pharmacological class or molecular structure, but instead can be qualitatively and quantitatively altered by seemingly subtle influences that are environmental or behavioral in origin. Well in keeping with clinical experience, experimental studies have shown that the behavioral effects of drugs can differ widely across individuals or within the same individual when administered under different conditions. The type of behavior, the context in which it occurs, the events controlling behavior, and the prior experience of an organism have all been shown to be critical determinants of the behavioral effects of drugs [Kelleher and Morse, 1968; McKearney and Barrett, 1978; Barrett, 19871. Moreover, comparable behavioral effects may be produced by administration of a host of compounds having quite different molecular pharmacologies. Thus, although not denying that the ultimate substrate of drug action involves the initiation and propagation of biochemical and biophysical events, behavioral pharmacologists realize the major importance, and oftentimes overwhelming influence, of behavioral or environmental variables as determinants of the behavioral effects of drugs.

There is a diversity of theoretical structure and a multiplicity of interpretations in behavioral pharmacology that often exceeds that found in other areas of experimental pharmacology. This may be a consequence of the wide range of observations that behavioral pharmacologists study and the wide berth and multiplicity present in descriptions of behavior. The effects of drugs on elicited responses, schedule-controlled behaviors, schedule-induced behaviors, and naturally occurring responses such as feeding and aggressive behavior, as well as complex human verbal reports, all fall within the domain of behavioral pharmacology. Analyses of behavioral effects of drugs have been made in terms of formal or structural features of behavior, presumed underlying hypothetical controlling variables, directly observable variables, behavioral processes, and constructs from various other fields of pharmacology and psychology.

Given the diversity and, at times, the appearance of disorder in the behavioral effects of drugs, it is important for the behavioral pharmacologist to provide, when possible, either generalizations that coordinate diverse findings, integrations of various effects, or general principles that organize or “explain” findings. In the first systematic textbook on behavioral pharmacology, Thompson and Schuster [1968] state that “one of the principal aims of pharmacology is to determine the mechanisms of action by which a drug produces a given effect; and this is swiftly emerging as the principal goal of behavioral pharmacology.” Carlton [1984] notes that “in pharmacology, the process of analysis eventuates in the isolation of physiological mechanisms of action-a statement about how certain drugs work so as to produce their characteristic diversity of effect. The same process is also a vital part of the subdiscipline of pharmacology called psychopharmacology.”

A large proportion of the work of behavioral pharmacologists is currently devoted to the study of basic pharmacology using behavior as a biological endpoint. With behavior of intact organisms, behavioral pharmacologists can study basic questions of mechanism as a pharmacologist with an isolated tissue preparation. Here, the questions asked, and the conclusions rendered are often generically similar and are usually directed toward molecular analysis. Do two compounds produce common behavioral or tissue changes by interacting with the same receptors or by inducing similar intermediary biochemical events? These experiments typically address problems related to potency, efficacy, and antagonism. Such work is an integral part of behavioral pharmacology and many impressive contributions have come from this type of

Analysis of Behavioral Effects of Drugs 391

inquiry. In addition, such work is essential to clarifying our basic understanding of drug action and in extending the utility of concepts from the tissue bath or homogenate to the integrated level of behavior. No mntter how promising the biochemical or molecular pharamacological profile of a new compound, the promise of therapeutic efficacy cannot be fulfilled until the drug is off the bench and into the organism.

However, behavioral pharmacology has an as yet unexpressed role to play in shaping advances in neurobiology , the experimental analysis of behavior, and drug development. The unique contribution of behavioral pharmacology to these fields may be based on its attempts at elucidating general principles that organize or “explain” the behavioral effects of drugs. In neurobiology, the finding of selectivity of drug action on a behavior or behavioral process would provide unique opportunities to discover underlying neural mechanisms. Without appropriate guidance, neurobiologists might search for neural correlates of behavioral phenom- enon which are not related to any specific behavior or behavioral process. Drug discovery programs currently involved with psychiatric compounds often do not have effective models for screening compounds. Detailed understanding of the basic principles underlying behavioral effects of drugs may help in rational drug discovery through development of models with predictive and functional validity.

In the present paper we discuss approaches to the study of the behavioral effects of drugs that fall under two major categories: (1) description and analysis of behavioral effects of drugs, and (2) behavioral mechanisms of drug action. The first includes studies that are primarily descriptive and examine behavioral and environmental factors that influence the behavioral effects of drugs. Discussion in this section focuses on the virtues and problems associated with analytic methods for obtaining information necessary for the derivation of general princples of drug action. The second category includes studies that attempt to assess the behavioral mechanisms by which drugs exert their effects. The distinction between these two categories is at once real and artificial. The complete analysis of factors that influence the behavioral effects of drugs may eventuate in an explanation of those influences involving proposed mechanisms.

Description of behavioral effects of drugs is a necessary first step toward understanding the manner in which drugs affect behavior. For example, Figure 1 shows that three classes of drugs generally produce comparable effects on nonpunished responding maintained by food presentation and on punished responding maintained by shock escape (Fig. 1, solid arrows at center). By contrast, these compounds have opposite effects on responding maintained by shock escape and on punished responding maintained by food presentation (Fig. 1 , dotted arrows at center). Systematic trends or orderly sequences in behavioral effects of drugs such as these are important in the development of more encompassing statements. These generalizations may ultimately lead to the understanding of behavioral mechanisms of action. What was once descriptive may evolve into a mechanistic account. Such integration is necessary for behavioral pharmacology to more readily incorporate and embrace data from other fields, as well as to have a more influential impact on related sciences.

DESCRIPTION AND ANALYSIS OF BEHAVIORAL EFFECTS OF DRUGS

In an influential paper, Dews [ 19581 suggested several environmental variables which may modify the behavioral effects of drugs. One unique contribution of the paper was due, to a large extent, to the analysis of the effects of psychoactive drugs in terms borrowed from behavioral psychology. In so doing, Dews, as well as other behavioral pharmacologists, brought to the study of the behavioral effects of drugs the wealth of knowledge from the experimental analysis of behavior, and the analytical skills accrued from exacting behavioral techniques. As a natural consequence of the application of these behavioral methods, experimental approaches to the interpretation of the behavioral effect of drugs were dramatically altered. Before the application of behavioral analysis, most studies of the behavioral effects of

392 Witkin and Katz

REINFORCEMENT

PUNISHMENT

FOOD PRESENTATION SHOCK ESCAPE

Amphetamines ( t ) Minor Tranquilizers ( t ) Opiates ( i )

‘r i

Amphetamines (+) Minor Tranquilizers ( t ) Opiates (1)

Amphetamines ( t ) Minor Tranquilizers (+ ) Opiates ( t ) f

\ Amphetamines ( t ) Minor Tranquilizers ( t ) Opiates (+ )

Fig. 1 . Effects of drugs on responding maintained under fixed-interval schedules of either food presentation or termination of a stimulus correlated with electric shock delivery (shock escape). Upright arrows indicate the occurance of increases in rates of responding across part of the dose-response function, whereas inverted arrows indicate that no rate increases were observed. Note the symetrical and asymetrical relationship inherent in the qualitative effects of these compounds (diagonal panels). (Re- produced from Barrett [1987] with permission.)

drugs, as well as most psychological studies, used mentalistic or post hoc explanations of behavior that had little power beyond their everyday familiarity. Drug effects were interpreted in terms of their actions on hypothetical and often ill-defined controlling variables (e.g., motivation, emotion). The application of behavioral principles allowed for an experimental analysis of the effects of drugs.

This section discusses methods for study and accounts of the behavioral effects of drugs used as alternatives to traditional analyses. These techniques form a diverse, but interrelated, series of experimental tactics that both individually and as a whole provide important information on the pharmacological modification of behavior.

Rate of Responding as a Determinant of Behavioral Effects of Drugs

There is now a large body of data that indicates that the rates and patterns of responding maintained prior to drug administration can be important influences on the behavioral effects of drugs. Those data have been reviewed [Robbins, 1981; Sanger and Blackman, 19761 and discussed [Thompson et al., 19811 at length elsewhere. These studies have indicated that the behavioral effects of drugs can be influenced quite dramatically by the rates of responding that are maintained. Most often, low rates of responding are increased by many drugs, in particular psychomotor stimulants, to a proportionally greater extent than higher rates of responding. Furthermore, there is often a linear relation between the effect of the drug and the control response rate [cf. Dews and Wenger, 19771.

If it were not for the large body of information on rate-dependent effects of drugs, it might seem counterintuitive that behavioral pharmacologists would appeal to this variable as a determinant of the behavioral effects of drugs. Rates and patterns of responding are descriptions of the form or structure of behavior-a notoriously poor subject for study. The experimental analysis of behavior has routinely looked beyond the mere form of behavior, and toward a functional analysis of its controlling variables. Since formal similarity does not imply that the determinants of the behaviors are similar, reliance on formal similarity to infer functional similarity has been termed the Formalistic Fallacy [Skinner, 1969, p. 89fl. In an analysis of responding maintained by shock avoidance, Sidman [ 19631 expressed a similar concern for functional analysis: “If a drug acts through a variable as general as the animal’s


interresponse times, it cannot be classified as one that specifically affects avoidance behavior.”

The importance of a functional analysis in behavioral pharmacology is emphasized by the results of a study by Kelleher et al. [1961]. In that experiment, responding of rats was maintained under fixed-interval schedules of food presentation. In one group of rats, the response was a downward depression of a lever; the other group of rats was trained to press a wall-mounted disk. Pressing the lever was typically accomplished with the forepaws while depression of the disc was generally with the nose. Although the responses were formally or topographically dissimilar, they were both under functional control of the fixed-interval schedule as evidenced by the development of the characteristic temporal patterns of responding. Further, the effects of d-amphetamine and meprobamate on the two behaviors were similar. Thus, the two responses, while formally dissimilar, were under similar functional control by the schedule of food delivery and were comparably affected by the drugs.

A converse situation which illustrates similar points is exemplified in an experiment by Teitlebaum and Derks [1958], in which licking by rats on a water tube postponed electric shock. Amphetamine increased the rate of licking. By contrast, the formally similar licking responses induced by injections of hypertonic saline, were decreased in rate by amphetamine. Thus, two formally similar responses were functionally dissimilar in terms of the controlling variables and the differential effects of amphetamine.

Thus, in predicting the effects of drugs on behavior, it is important to determine as best as possible the functional control over the behavior. Behaviors that appear formally similar may be affected by drugs in quite a different manner if functionally dissimilar. Alternatively, behaviors that appear formally dissimilar may be affected by drugs in a similar manner if functionally similar.

Although a functional analysis of behavior may be of great significance in the determination of drug effects, there is no reason to conclude that structural features of behavior are not important. For example, topography of response may influence the maximal rate of responding that can be produced by a drug. Rate, topography of behavior, and other structural aspects of behavior may become more or less influential under certain conditions (see next section below) or with particular responses [cf. Graeff and de Oliveira, 19751.

Because of the emphasis on rate of responding as a determinant of drug effect, studies investigating the modification of drug effects by other variables have quite appropriately controlled for response rate differences when evaluating the influence of a variable on the behavioral effects of drugs. For example, McKearney [1974] and Barrett [I9761 examined the effects of several drugs on performances maintained by either food or electric shock presentation. These studies indicated that the event maintaining responding can dramatically influence the effects of certain drugs, and dispelled suggestions that the event that maintained behavior was a relatively unimportant factor in determinng the behavioral effects of drugs [for a review, see Barrett and Katz, 19811.

Given a functional analysis, it should have been no surprise when the effects of drugs were found to depend on the type of event that maintained the behavior. As reported by several investigators [McKearney and Barrett, 1978; Morse et al., 19771, behaviors maintained by food presentation or electric-shock presentation are a function of several overlapping and nonoverlapping sets of variables. As Skinner pointed out years ago:

Some people are primarily under aversive control. They are acting to avoid trouble or as if to avoid trouble when the present situation may be quite harmless. These are the compulsive, the threatened. . . . On the other hand, some more fortunate individuals are behaving mainly under positive reinforcement. The world has been good to them; it has paid off in many ways. They find themselves as active as the compulsive, but for quite di#erent reasons. It would be amazing if a drug had the same effect upon individuals who differ as greatly as this in terms of personal

394 Witkin and Katz

I D

CCNTRCC RESPONSE RATE

Fig. 2 . Hypothetical functions depicting possible relationships between the control rate of responding maintained by different events (0, 0 ) and the effects of drugs. On the basis of experimental data, it is assumed here that the drug produces differential effects on comparable response rates at point X (abscissa) and that this represents an intermediate rate value. The dashed line at 100% represents control, or nondrug, rates of responding; points above and below this line represent drug-induced increases and decreases, respectively. None of the relationships shown reflects an invariant relationship (i.e., have no slope) between response rate and drug effects. Although an outcome of this type is possible, it appears to be characteristic of low doses that are not typically behaviorally active. Similar drug effects across events are obtained when control rates are high (A), low (B), or at both high and low values (D); (C) similar effects are obtained when response rates maintained by one event are low (y) and those maintained by a different event (x) are high. Reproduced from Barrett and Katz [1981] with permission.

histories. Until we can understand the functional relations governing behavior, the effect of a drug upon an arbitrary measure, no matter how quantifiable, may be puzzling and misleading. [Skinner, 1959, p. 227 (emphasis added)].

Results of studies indicating that the event that maintains behavior can be an important determinant of the behavioral effects of drugs have been interpreted as indicating that rate dependency can be applied to a more limited range of circumstances than was originally proposed. However, these studies were not designed in a manner to allow an assessment of the relationship of drug effect to control response rate; the effects of drugs were assessed on responding maintained at comparable rates by different events in order to eliminate rate as a potential factor contributing to differences in drug effects. The rate-dependent effects of the drugs could not be determined because the effects of the drugs across different rates of responding were not assessed. The matching of response rates maintained by the different events allows for the assessment of the effects of drugs on rates of responding at only one point on the rate-dependency function [cf. Barrett and Katz, 19811. For example, responding maintained by food may be affected in a rate-dependent manner that is different from that for responding maintained by electric shock (Fig. 2). As illustrated in Figure 2 , delineation and comparison of specific rate-dependency functions is required for a more complete interpretation of the contribution of response rate as well as other variables that may modulate the behavioral effects of drugs [cf. Barrett and Katz, 19811.


0000

1000

100.

10 ! . ' ' ' "" I ' . ' ' ""I . ' ' ' "1.1

.001 .01 .1 1

Control Response Rate (Responses/sec)

Fig. 3. Effects of methaqualone on punished behavior are modulated by the punishing shock intensity. Methaqualone either increases punished responding to a greater or lesser extent than nonpunished responding (0 mA), depending on the shock intensity. Each point represents effects of 30 mg/kg methaqualone in a single pigeon as a function of the control rate of responding when no drug was administered. Coordinates are logarithmic. Lines were fit to the data by linear regression analysis. (Reproduced from J.M. Witkin, J.L. Katz, and J.E. Barrett, unpublished observations.)

Specificity of Drug Action

Historically, behavioral pharmacology has been guided in large part by an attempt to pinpoint specific behaviors or behavioral processes that are affected by drugs. Specificity of drug action is directly translatable into experiment. For example, does a drug increase punished behavior at doses that do not increase other nonpunished behaviors? Wuttke and Kelleher [ 19701 reported that comparable rates of punished and nonpunished responding were affected similarly by benzodiazepines. Others have provided data showing that punished responding is increased to a greater extent than comparable rates of nonpunished behavior [Barrett and Jeffrey, 1979; McMillan, 19731. It is clear from the literature that a specific action of drugs on punished behavior has yet to be convincingly demonstrated by these methods. However, with a compilation of results came a better understanding of some of the Variables influencing behavioral effects of drugs, and importantly, the appreciation that precise control over these variables is required to adequately address the problem of the behavioral specificity or selectivity of drug effects.

Our understanding of the variables that control behavior suggests that behavioral processes, even as seemingly straightforward and unencumbered as punishment, are complexly determined. For example, one determinant of punishment is the intensity of the punishing stimulus. Manipulation of this variable can dramatically influence the behavioral effects of drugs. Figure 3 shows that the relationship between drug effects and control rates of punished behavior can be modulated by the intensity of the punisher. Responding of pigeons was maintained under a 5-min fixed-interval schedule of food presentation. In separate experiments, every 30th response during the fixed-interval produced either a 0-, 2- , or 4-mA pulse of electric shock. A twofold change in shock intensity from 2 to 4 mA did not significantly alter overall rates of suppressed responding [Witkin and Barrett, 1976; Witkin et al., 19811. However, 30 mg/kg methaqualone increased rates of punished responding to a greater extent than comparable rates of nonpunished responding only at the lower shock intensity. At the higher intensity, punished and nonpunished responding were affected similarly. Thus, the

396 Witkin and Katz

800 7

- e g 600- 0

8 3 400- m r 0

C

c

I

v

2 200- 0 r

0 -

Chlordiazepoxide (mg/kg)

Fig. 4. Food presentation modulates the increases in punished responding produced by chlordiazepoxide hut is not necessary for this behavioral effect. Responding of rats (n = 6) produced either food plus electric shock or shock alone. (Reproduced from J.M. Witkin and L.A. Perez, unpublished observations.)

entire question of specificity of drug action can be dependent on basic variables controlling behavior.

Another approach to defining specificity of drug action is through documentation of the generality of the behavioral effects of a drug. If a drug selectively affects punished behavior, increases in punished behavior should be obtained over a wide range of conditions (e.g., the event maintaining responding, the type of punisher). However, by these methods also, initial hopes of finding a specific effect of a drug on punished responding have been mitigated by the subtle, yet powerful, influence of the variables controlling behavior and drug action [cf. Branch et al., 1977; de Carvalho et al., 1981; McKearney, 19761. Thus, the problem of determining specificity of drug action is not as straightforward as was initially hoped. Drug effects are highly malleable. Delineation of a selective behavioral drug effect is thus dependent on adequate characterization of the necessary and sufficient conditions of a particular behavioral effect.

Determination of the Necessary and Sufficient Conditions for a Drug Effect

Information on the necessary and sufficient conditions for a drug effect is fundamental to defining the range of conditions under which specificity of action applies. These data can also assist in narrowing the range of potential mechanisms of action of drugs [see Cook and Davidson, 1973; Kelleher and Morse, 1964, for discussions of punished behavior].

For example, as many of the drugs that increase punished responding also increase food and fluid intake, a reasonable question has been the relationship of this pharmacological action to the increases in punished behavior [cf. Leander, 19831. The complexity of behavioral drug effects is illustrated by experiments that clearly implicate food presentation as a modulator of the effects of drugs on punished behavior. Figure 4 presents results of an experiment that compared the effects of chlordiazepoxide on food-maintained responding suppressed by response-produced shock delivery with responding suppressed by shock presentation where no food was delivered. In this experiment, barpress responses of rats were maintained under a multiple schedule in which every 30th response in one component produced food. In an alternate component, every 10th response produced either a food pellet plus electric shock or shock alone. Comparable rates and patterns of responding were maintained across conditions. Although chlordiazepoxide increased responding in both cases, the increases were larger


600 - 500 -

400 -

300 -

200 -

100 -

f

c 0

e c 0 ;;e I

0 c

t

:: d

m C

Electric-Shock Avoidance Yoked Electric-Shock Presentation

1

T r Y C 0.03 0.1 0.3 1 .o

6-Amphetamine

8 rn

T I

C 0.03 0.1 0.3 1 .o

Fig. 5 . &Amphetamine increases punished responding of squirrel monkeys after exposure to a shock avoidance schedule but not after presentation of shock per se. 0, Drug effects prior to avoidance or yoked shock presentation; 0, drug effects after exposure to shock schedules. (Adapted from J.E. Barrett and J.M. Witkin, 1986, with permission.)

in the classic punishment condition in which responding also produced food. Thus, the results of these experiments indicate that food presentation can markedly alter the behavioral effects of chlordiazepoxide. At the same time, food presentation is not a necessary condition for drug-induced increases in suppressed behavior.

A further illustration of the need for detailed behavioral analysis in the understanding of the mechanism of action of drugs is shown by experiments on the role of behavioral history in altering the effects of drugs on punished responding [Barrett and Witkin, 19861. Barrett [1977] demonstrated that the typical effects of d-amphetamine could be completely changed by providing squirrel monkeys with certain behavioral experiences. In particular, although d- amphetamine normally decreases punished responding, increases in punished responding were obtained in monkeys given prior exposure to conditions in which responding postponed electric shock delivery (avoidance). In the absence of drugs, there were no apparent differences in punished behavior between those animals with, and those without an avoidance history. Thus, the qualitative modification of the behavioral effects of d-amphetamine likely resided in the avoidance history, a remote experience. Behavioral experience, although not overtly reflected in current behavior, can make itself manifest upon drug challenge. With the avoidance history, d-amphetamine presents a profile of behavioral effects that are typical of anxiolytic agents.

In order to evaluate whether exposure to electric shock was sufficient for the increases in punished responding, squirrel monkeys were exposed to either a shock avoidance schedule or were given response-independent shock with the same frequency and temporal pattern as the animals responding under the avoidance schedule [Barrett and Witkin, 19861. Results of this experiment indicated that presentation of electric shock per se was not sufficient to endow d-amphetamine with the ability to increase punished behavior (Fig. 5) . Thus, factors related to the avoidance behavior appeared to be critical to the qualitative changes in the effects of d-amphetamine; more general factors related to shock, such as stress, did not appear to play a major role. One clue as to the factors possibly contributing to the influence of the avoidance history was suggested by data obtained on the duration of responding. Whereas exposure to the avoidance schedule decreased response duration 40%, no change was observed in monkeys exposed to response-independent shock [Barrett and Witkin, 19861. The avoidance history may therefore induce long-term changes in features of the response that interact with subse- quent drug administration in a distinct manner. Such a possibility, in its extreme form,

398 Witkin and Katz

160-

140-

::I: 80 - 60 - 40 - 20 - 0-

11 -

500 - 450 - 400 - 350 - 300 - 250 - 200 - 150-

100-

50 - 0-

C

7

C

0.01 0.03 0.1 0.3 1 .o d-Amphetamine (mgkg)

3 5.6 10 30.0 56.0

Chlordiazepoxide (mglkg)

Fig. 6. Effects of d-amphetamine and chlordiazepoxide on responding of squirrel monkeys maintained under a multiple shedule of electric shock postponement and shock punishment. Monkeys had previously been exposed to response-independent electric shock presentation. Data represent mean effects in two monkeys. (Adapted from J.M. Witkin and L.A. Dykstra, unpublished observations.)

suggests a locus for the changed behavioral effect of d-amphetamine in specific characteristics of the response (as with rate dependency). Empirically, this could be evaluated by inducing comparable topographical changes without the avoidance history or by providing experience under avoidance schedules which do not alter response topography in the same way.

In another experiment, animals previously exposed to schedules of response-independent shock were later given experience under shock-postponement schedules [J.M. Witkin and L.A. Dykstra, unpublished observations]. After exposure to response-independent shock, responding of squirrel monkeys was maintained under a multiple schedule consisting of al- ternating components of shock avoidance and punishment. McKearney and Barrett [ 19751 reported increases in punished responding with d-amphetamine under this schedule. Figure 6 shows that the history of response-independent shock overshadowed the influence of the shock-avoidance experience; d-amphetamine did not increase punished responding in monkeys previously exposed to response-independent shock. Although d-amphetamine did not increase punished responding, increases in avoidance responding were obtained with d-amphetamine. Thus, the response-independent shock did not simply eliminate the rate-increasing effects of d-amphetamine. Likewise, the ability of chlordiazepoxide to increase punished responding


was not affected by prior exposure to electric shock indicating that the punishment baseline was not insensitive to the response rate-increasing effects of drugs. In addition to defining another type of behavioral history which can alter the behavioral effects of drugs, the data presented in Figure 6 provide an additional leverage point for delineating the necessary and sufficient conditions for increases in punished responding with &amphetamine.

Clinical Correlational Approaches

Analysis of behavioral effects of drugs is often directed toward an understanding of their clinical utility. One classic approach has been to develop models which predict therapeutic efficacy. Punished behavior has been used in this way to model anxiolytic drug responses for drug development. Anxiolytic and sedative-hypnotic compounds increase punished behavior whereas a host of compounds from other pharmacological classes do not produce this behavioral effect [cf. Cook and Davidson, 19731. In order to detect new compounds with related therapeutic capability, the model need not be functionally equivalent to the clinical response, only predictive of it.

Likewise, models with predictive utility need not have face validity. A model of anxiety with face validity would have a laboratory animal engage in responses which resemble the anxiety symptoms in humans (e.g., autonomic arousal, muscle tension, hypervigilance). One such model might be that of an organism responding under a schedule of electric shock postponement. However, anxiolytic drugs do not affect responding under these baselines with any specificity. Thus, structural models do not predict clinical efficacy; they merely look good. By contrast, models with predictive value need not be concerned about appearances, only that they predict clinical responses.

Carlton [1983, 19841 has discussed the application of model testing to the assessment of mechanisms of drug action. The adequacy of the model for this purpose can be evaluated by precise correlational and experimental methods some of which have already been described. Briefly, do parameters of drug action in the model correspond to the clinical response? In an adequate functional model of anxiety, drug potencies in increasing punished responding should correlate with drug potencies in the treatment of anxiety. Other aspects of the clinical response to anxiolytics should also correlate with responses in the model. These clinical features of the pharmacological response include such factors as tolerance development and the duration of drug treatment required for therapeutic efficacy. According to this analysis, the more correlations that can be discovered between the model and the clinical response, the greater is our confidence that the two behaviors are under similar functional control; functional relationships relating drug effects in the model are comparable to those governing the clinical response.

Other mechanisms of drug action related to therapeutic efficacy can then be evaluated by these methods. If it is postulated that anxiolytic agents are effective because of their ability to affect neuronal firing rates of dorsal raphe neurons, correlations of electrical responses and clinical effects should also be strong for relevant parameters of drug action such as potency, time-course and tolerance development. Furthermore, if punished responding is a functional model of clinical response to antianxiety agents, the correlation of drug effects on electro- physiology and punished behavior should also be strong.

Functional models may be central to future rational drug development. There currently exists a general paucity of adequate models of many psychiatric disorders. Behavioral methods for screening for novel anxiolytics, antidepressants, antipsychotics, and cognitive-enhancers are imperfect. An important contribution to our future understanding of such disorders and novel drug discovery may depend upon increased efforts in the experimental analysis of behavior to fully characterize relevant functional relationships. With the use of functional models comes the possibility of elucidating mechanisms of drug action. For example, do increases in punished behavior represent an effect on the punishment process or on behaviors that collectively could be summarized as anxiety? Such a question is compelling, as an affirmative answer would lead directly to a better understanding of these processes. A com-

400 Witkin and Katz

pound that affects anxiety directly implies a definable neurobiological correlate of anxiety. Functional models of anxiety could then be used to characterize these neurobiological processes, further understanding of which in turn may lead to refinement in the compounds with behavioral specificity and in the behavioral models [e.g. Mansbach et al., 1988; Witkin et al., 1987 and other papers in this issue].

Interaction of Drugs with Behavioral Variables

Sidman, writing in 1956 about the accelerating pace in behavioral pharmacology, merely a year after its first seminal paper [Dews, 19551, wrote that the field “faces the formidable task of systematizing the observed relations into an empirically sound and rational classification.” In an attempt at such a scheme, Sidman [1956, 19631 described the analysis of behavioral effects of drugs as the investigation of the interaction of drugs and the variables of which behavior is a function. To proceed properly in this endeavor, a thorough knowledge of these controlling variables is required in full parametric detail. Dealing with avoidance responding, for example, does the drug affect the relationship between avoidance behavior and the shock-shock interval, the response-shock interval, or both? Does it act by altering the functions relating shock intensity to rate of avoidance responding? Determination of dose- effect functions at each of a range of parameter values of each controlling variable would then provide indications of the drug affecting behavior through an interaction with a controlling variable. If a drug produces, for example, a parallel shift in the shock intensity function but does not alter the other functions, it might be that avoidance behavior is affected by a drug action, in the manner that shock intensity controls responding.

This is an intriguing notion but is beset by a host of problems. Sidman noted that the range of relationships that needed to be explored was phenomenal, requiring great patience and experimental ingenuity and vigor. There is, 30 years after this suggestion, a relative scarcity of data along these lines.

An additional complication in the analysis of behavioral variables as targets of drug action is that the effects of drugs on behavior involve interactive processes. Drug effects themselves are altered by the specific conditions under which they are studied and by the behavioral effects they produce (see discussion of drug-event and drug-behavior interactions in the next section).

BEHAVIORAL MECHANISMS OF DRUG ACTION

In the initial conception of behavioral mechanism of drug action, Dews [1956; see also Dews, 1958, 19701 suggested that if a drug had a particular behavioral mechanism of action, the effects of the drug should be the same as the effects of an appropriate independent variable. With a few exceptions, later publications addressing behavioral mechanism of drug action have neither assessed similarity of drug effects and other behavioral variables, nor explicitly refined the meaning of behavioral mechanism of action. Thompson [ 19841 defined behavioral mechanism of action as a description of the effects of a drug on a particular behavioral system expressed in terms of some more general set of environmental principles regulating behavior. While this definition is explicit in several respects, it is different from the one originally implied by Dews [ 19561. Thus, included in the review by Thompson was a further discussion of the many factors that have been shown to influence the behavioral effects of drugs.

Demonstration that the effects of a drug are influenced by a certain independent variable, however, is different from demonstrating that the effects of the drug are similar to the effects of some independent variable. For example, Laties and Weiss [ 19661 showed differences in the effects of several drugs on responding maintained under fixed-interval schedules with and without an added “clock” discriminative stimulus. The difference in effects under the two procedures showed that the effects of the drug were modulated by the differences in the two procedures, however, it did not indicate that the effects of the drug were similar to effects of


some independent variable that changed stimulus control. Results such as these may only suggest where to look for behavioral mechanisms of action.

Conclusions that the effects of a drug are influenced by some variable, or that the administration of drug is similar to some environmental change, are different in kind. Any variable may operate to influence drug effects without actual involvement in the mechanisms of action. For example, the clock stimulus in the study by Laties and Weiss may have altered drug effect indirectly through modification in the rate or pattern of responding, rather than by an action on a common set of operations and behaviors collectively termed the stimulus control of behavior.

For behavioral mechanism of action to be a useful concept in the analysis of the behavioral effects of drugs, it must be distinguished from the hypothetical mechanisms that are often supposed to explain the behavioral effects of drugs. An example of a hypothetical mechanism that has been the spark for several current lines of research in behavioral pharmacology is the notion that drugs alter motivational states that are in turn responsible for behavior.

One view of anxiolytic drug action, for example, is that their behavioral effects are an outcome of their effects on anxiety. Dantzer [ 19781 argued that if benzodiazepines act by reducing anxiety, fear, or frustration, then several behavioral outcomes should occur. Since both punished responding and the the facilitation of avoidance responding by a stimulus terminating in response-independent shock are presumably controlled by anxiety, benzodiazepines should decrease the effectiveness of both the punishing and pre-shock stimuli. How- ever, only the former prediction was borne out by an experiment implying that either the initial assumptions concerning controlling variables, the mechanisms by which benzodiazepines act, or both were incorrect.

The distinction between hypothetical mechanisms and behavioral mechanisms is essen- tially similar to the distinction between hypothetical constructs and intervening variables. According to MacCorquodale and Meehl [1948], an intervening variable is an abstract re- statement of empirical relationships, whereas the defining characteristic of a hypothetical construct is the supposition of unobservable process or entities. In the case of underlying motivational states, the construct includes words (e.g., anxiety, hunger) that are not explicitly defined by the empirical relations under study. Therefore, evidence of the empirical relations is not necessarily confirmation of the construct because the terms of the construct are not defined by the empirical relations. In the study by Dantzer [1978], the observation that the drug increased punished behavior was not sufficient to indicate that the drug induced an alteration in anxiety. Anxiety was not defined by an empirical relation to the set of terms describing punished behavior.

A behavioral mechanism of action is similar to an intervening variable in that the terms used in defining the mechanism are explicitly defined by the empirical relations. For example, if a behavioral mechanism of action for a drug is a decrease in stimulus control of behavior, then administering the drug should produce specifiable effects that are determined by the meaning of the term stimulus control. In this case, the proof of the empirical laws constitutes the necessary and sufficient conditions for the proof of the abstraction; there are no terms that are not explicitly defined by the empirical relations.

Had the experiment by Dantzer [ 19781 confirmed his hypothesis, the intervening anxiety would have heuristic value as a construct providing organization or simplicity. A wide range of facts encompassed into a small set of principles can render constructs useful as descriptive terms and in the facilitation of the science [Marr, 19903. Nonetheless, such constructs, lacking explicit empirical definition, retain the inherent problems of hypothetical constructs.

A description of drug effects in terms of behavioral mechanisms of action does not imply that the sequence of biological events occurring with drug administration is unimportant or unrelated to the behavioral effects observed. It is often appealing to suggest that these are events taking place “at some other level”; however, the real distinction is that the linkage of

402 Witkin and Katz

these biological immediate causes to their behavioral outcomes in unknown. The relation between behavioral and neurobiological effects of drugs is complex and likely involves re- ciprocal interactive dynamics [see Barrett and Tessel, 1974; Seiden et al., 1975, and other papers in this issue]. A full mechanistic account of a particular behavioral effect of a drug would describe completely the role of each of these processes. Those unsettled with behavioral action at a distance [Branch and Schaal, 19901 may seek “explanation” of the behavioral effects of drugs through appeal to these more “immediate causes”. Behavioral mechanisms of action as intervening variables, however, serve to emphasize the important dynamics between drug effects and their behavioral and environmental determinants.

Proof of Empirical Laws

What constitutes proof of behavioral mechanisms of action has not been given extensive consideration. Dews [1956] suggested that the analysis of the behavioral effects of drugs should proceed with a determination of the extent to which the drug effects are like, in the sense of having the same effect on behavior, the effects of the independent variables. For example, a possible mechanism for a drug-induced change in rates of responding is a decrease in the control of behavior by discriminative stimuli. A drug having this as its exclusive mechanism would have effects on schedule-controlled behavior similar, in all manifestations, to the effects of other variables that decrease stimulus control; specifically, decreases in the differences between the discriminative stimuli. Thus, as suggested by Dews [1958, 19701, to test whether a drug has a particular mechanism, one must assess whether performances after drug administration are similar in form to those after the appropriate environmental manipulation.

Moreover, determining whether the effect of the drug and the effect of the environmental manipulation are similar in form can often be ambiguous. First, the formal equivalence of behavior following a change in an independent variable and administration of a drug can be difficult to directly quantify. Often there is no appropriate metric with which to scale the formal similarities of behaviors taking place within an experimental space over time. More difficult an issue, however, is that, as noted above, the Formalistic Fallacy indicates that similar behaviors may be functionally dissimilar [Skinner, 19691. For example, in the presence of a houselight under a conditional discrimination procedure, a response on a red key was reinforced according to a fixed-interval schedule (SD response), whereas a response on an amber key was the SD response in the absence of the houselight. Over a range of conditions, the performances under this procedure were characterized by a high degree of stimulus control [Katz, 19821. Under other conditions, (e.g., decreased houselight intensity), the degree of stimulus control was reduced [Katz, 19831. Adding a requirement that an SD response was ineffective until a fixed number of SA responses were emitted (conjunctive schedule), produced effects on rates and patterns of responding that were similar to those of decreasing the intensity of the houselight. These two operations, however, were functionally different, as was revealed when drugs were administered [Katz, 19831.

Since formally similar effects of an environmental change may be attributable to functionally dissimilar mechanisms, it was suggested that the term behavioral mechanism of action be reserved for those instances in which it can be determined that the administration of the drug is functionally equivalent to changes in that independent variable [Katz, 19901. A definition requiring functional equivalence excludes mere formal equivalence, as well as instances in which some variable influences the manifestation of the drug effect.

It may be especially difficult to assess whether the effects of drugs are functionally equivalent to the effects of some other independent variable. Drugs typically have more than one effect; as such, a drug effect that is functionally equivalent to some environmental manipulation may be occurring along with other drug effects. These auxiliary effects may mask the functional equivalence. For example, when drugs function as reinforcing stimuli, rates of responding at high doses per injection may be lower than rates of responding that occur at


lower doses. The lower rates of responding that are maintained are often attributed to effects, other than reinforcing effects, that occur due to accumulated injections of the drug. These side effects disrupt ongoing operant behavior and thus mask the reinforcing function of the drug [Woods et al., 1987; Katz, 19891. The behavior maintained by high doses of the drug may not formally resemble responding maintained by nonpharmacological reinforcers, and therefore its functional equivalence to behaviors maintained by other reinforcers may not be apparent. Thus, a drug may have an effect through a specific behavioral mechanism (in this case, a reinforcing function), as well as other effects on behavior; those other effects may obscure formal equivalence.

Methods of Determining Functional Equivalence

One obvious method of determining whether administration of a drug is functionally equivalent to altering some non-pharmacological independent variable is to compare the relations of the behavioral dependent variable to either drug dose or value of independent variable. However, since a drug may have effects in addition to its behavioral mechanism of action (as described above), functional relations between dose and dependent variable may be influenced by these other effects. As a consequence, dose-effect relationships may be dissimilar to relations between the non-pharmacological independent variable and behavior. The assessment of functional equivalence of two non-pharmacological operations may be similarly problematic. Different, ostensibly equivalent operxions have been shown to have varied functional relations between changes in parameter and change in particular behaviors [cf. Miller, 19591. Since it seems unlikely that any change in independent variable will be singular in its effects, it appears that a comparison of functional relations will not suffice in determining the functional equivalence of two operations.

One method that has been used to assess the functional equivalence of administration of drug and operations that attenuate stimulus control, is to determine if the drug effects are blunted by an environmental change that increases stimulus control. This type of result has been referred to as an environmental antagonism of drug action [Katz, 19901. For example, under a fixed-interval conditional-discrimination procedure a response on a red key in the presence of a houselight was reinforced according to the fixed-interval schedule (SD response), whereas a response on an amber key was the SD response in the absence of the houselight. Each response randomly alternated the positions of the key colors (right or left) and the houselight on or off conditions alternated in a mixed sequence. Thus, the two sources of discriminative control over responding were the presence or absence of the houselight, and the key colors. Pentobarbital decreased stimulus control of responding at intermediate (3.0-10.0 mg/kg) doses that did not appreciably alter average rates of responding. d-Amphetamine decreased stimulus control only at high doses (3.0-5.6 mg/kg) that also substantially decreased response rates (Fig. 7, filled squares). When the houselight was rendered insignificant (houselight on and off conditions studied in isolation), the effects of the drugs on stimulus control were similar (Fig. 7 , open squares), suggesting that the two drugs primarily affected stimulus control exerted by the colors of the response keys.

Finally, the effects on stimulus control produced by the two drugs were studied without the random switching of key colors following each response, thereby adding position to color of the key as a discriminative stimulus. This change increased the degree of stimulus control existing prior to drug administration and significantly attenuated the effects of the drugs on stimulus control (Fig. 7 , filled circles). Thus, an environmental operation that had effects opposite to those of the drug, attenuated the effects of the drug [Katz, 19881. Since the effects of the drug were attenuated by an operation that increased stimulus control (environmental antagonism), the decrement in drug effect is evidence for stimulus control as a behavioral mechanism of drug action. Similarly, environmental changes that have the same effect as the drug should increase the effect of the drug. Assessment of whether environmental changes and

404 Witkin and Katz

Q Z -

W u - Z Q I - 0

0 -

0.05 -

-0.1 0- Mult w/o K L Swrtch

0 Slng’e C o m p o n e n t s

J -0.15’, , I , , , + J ~ I I I , , c 0.1 0.3 1.0 3.0 C 1.0 3.0 10.0 17.0

DRUG DOSE (MG/KG)

Fig. 7. Comparison of effects of &amphetamine and pentobarbital on stimulus control under several different experimental conditions. Abscissas: drug dose (in mgikg, log scale); ordinates: stimulus control (A’), w, Multiple fixed-interval schedule in which each response randomly alternated the position of the keylight colors and the status of the houselight randomly alternated during the timeout which followed each fixed-interval component; 0, single-component fixed-interval schedules (houselight-on and -off conditions studied in isolation) in which each response randomly alternated the position of the keylight colors; e, multiple fixed-interval schedule in which the position of the keylight colors and the status of the houselight randomly alternated during the timeout which followed each fixed-interval component. Note that the effects of each drug were similar under the multiple schedule (B) and single components (0);

furthermore, the effects on stimulus control were attenuated when the positions of the keylight colors did not alternate with each response (e). (Reproduced from Katz, 1990, with permission.)

drug administration interact in this manner appears at present to be the best way to determine whether those operations are functionally equivalent.

Behavioral Processes as Targets of Drug Action The term behavioral mechanism of action was reserved above for situations in which

administration of a drug is functionally equivalent to some environmental manipulation. A drug effect on a behavioral process, such as reinforcement or punishment, is a special case. In the following discussion, we concentrate on the finding [Branch et al., 19771 that the effects of pentobarbital on responding suppressed by electric shock can differ from its effects on responding suppressed by a stimulus during which responding has no scheduled consequences (timeout). The discussion, however, applies equally to many other results.

In the study by Branch et al. [1977] pentobarbital increased responding suppressed by electric shock but not timeout presentation. These results raise some interesting conceptional issues in the analysis of behavioral effects of drugs. Do different drug effects imply that there must be some observable or functional differences in the behaviors? Does it imply differences in the neurobiological correlates of the behaviors?

One possible explanation of these effects is that, while electric shock was functioning as a punisher, timeout was suppressing responding through some other mechanism. That timeout was functioning as a punisher was demonstrated by the observation that response-independent timeout presentations did not suppress responding; punishers are functionally defined by the behavioral effects they produce [Azrin and Holz, 19661.

Conceivably some parametric differences in timeout and shock may account for the differences in drug effect. By varying parameters of both shock and timeout (e.g., duration, schedule, intensity), opposite drug effects may be obtained. If after demonstrating that timeout-suppressed responding is not increased by drugs that increase shock-suppressed responding regardless of parameter, the conclusion that behavior suppressed by timeout or electric shock are different is more compelling.


One factor that may contribute to the differences in drug effects with the different punishers may be ancillary behavioral effects of the punishers. Elicited behaviors following electric shock have been well documented [Smith et al., 19721; however, there may also be ancillary effects of less obtrusive stimuli such as timeout. For example, the difference in effects described by Branch et al. [ 19771 could have been due to a reduction in responding for a brief period following timeout delivery but not after shock delivery. These effects may not become manifest until the drug effect enters into the myriad of variables controlling behavior and alters the dynamic interaction among these variables. The combination of drug and event constitutes a new environment which may affect behavior differently than either drug or event in isolation. Administration of pentobarbital, when the timeout or electric shock is removed, might then result in equivalent effects, regardless of punishing stimulus. Such an experiment would illustrate the concept of drug-event interactions in which the presence of an environmental event may alter the expression or manifestation of the drug effect (as in Figs. 3 and 4). The precise nature of the behavioral change induced by the drug-event interaction can be determined by experiment.

Drugevent interactions can be distinguished from drug-behavior interactions. In drug- behavior interactions, the drug induces a change in behavior which in turn alters some feature (e.g., frequency) of the environmental events controlling behavior. These changes produce further alterations in behavior as it adjusts to the environmental changes. For example, initial increases in rates of punished responding induced by pentobarbital could increase the rate of food delivery. The new rate of food delivery in turn may then act to further modify the behavioral effects of the drug (cf. Fig. 4). Drug-behavior interactions also allow for drug- event interactions to come into play once behavior has altered the presentation of events. Likewise, drug-event interactions may be responsible for initial differences in behavioral effects of drugs which subsequently trigger unique drug-behavior interactions. Clearly, the two types of interactions are highly interdependent. The search for specificity of drug action may be simplified by conducting studies which attempt to isolate the influence of these interactions in the observation of drug effects.

CONCLUSIONS

Historically, the effects of drugs on behavior have been interpreted in terms of hypothetical constructs. Over the last 30 years, behavioral pharmacology has instead emphasized an empirical approach to the analysis of the behavioral effects of drugs. Current interpretations of how drugs alter behavior are in terms of behavioral or environmental factors that influence drug effects or the behavioral mechanisms of drug action. Attempts are also made to relate these statements to neurobiological processes.

The discovery of new compounds with restricted pharmacological activities will un- doubtedly play a large role in helping to clarify our understanding of the behavioral targets of drug action. The novel anxiolytic, buspirone, for example, increases punished behavior but does not increase other behaviors [Barrett et al., 1988; Witkin et al., 19871. It seems unlikely that compounds will be discovered that selectively affect a behavior disorder as general, for instance, as anxiety. However, different variations of the disorder (e.g., panic, phobic) or their component behaviors, each probably under the control of a unique set of historical and current controlling variables, may be better suited to a reasonable specificity of pharmacological intervention. Elucidation of the variables of which these individual behaviors are a function and the roles they might play in drug action are the type of tasks for which behavioral analysis and behavioral pharmacology are well designed.

Behavior, even within the confines of the laboratory, is highly complex; it modifies and is modified by a multiplicity of subtle yet wieldy forces. Since drug effects on behavior are labile and subject to a variety of influences, the generality of determinants of the behavioral effects of drugs must come from a convergence of data from a series of experiments. The use

406 Witkin and Katz

of refined or highly analyzed behavioral systems [e.g., Schindler and Harvey, 1990, this volume], as with isolated tissue, may facilitate understanding of mechanistic questions. How- ever, the conditional nature of even general statements regarding drug action should not be discouraging; these data are the heart of behavioral pharmacology and do not differ funda- mentally from conditional statements of other disciplines.

In drug development, drug industries grapple with discovery of behaviorally active drugs of a variety of classes. However, with few exceptions, this is done without the rational guidance of a coherent behavioral theory. The development of novel anxiolytics, antipsychotics, antidepressants, and so-called cognitive enhancers provides cases in point. The determination of the efficacy of these compounds, in the final analysis, is a behavioral question. In neurobiology, research in many areas could be greatly aided by a refined and detailed account of the manner in which drugs affect behavior. The establishment of functionally specific behavioral effects of drugs or other manipulations (e.g., CNS lesion) will provide the refined behavioral preparations required for neurobiological inquiry into the regulation of those specific behaviors or behavioral processes. Such characterization falls within the domain of behavioral pharmacology. Issues in experimental psychology and psychiatry may also receive important clarification by information on strictly defined behavioral actions of drugs. More than 50 years ago, Skinner [1938] pointed out that the science of behavior could be considered a sort of “thermodynamics of the nervous system,” providing the widest descriptions of its activity. Valid laws of behavior cannot be proved wrong by studies of neurobiology; rather, behavioral studies impose limiting conditions on the science. The contribution of behavioral pharmacology is much the same, through its provision of a program of behavioral laws and descriptive principles, which can be employed for rational drug discovery as well as a frame- work with which the neurobiologist can search the proximal causes of functionally meaningful behavior.

ACKNOWLEDGMENTS

We thank Dr. J. E. Barrett for many hours of profitable discussion and for comments on an earlier version of this manuscript. The accommodations of Dr. and Mrs. D. J. Fisher, during which many of these ideas were formalized, are greatly appreciated. We are also grateful to Drs. M. N. Branch and M. J. Marr for providing us with copies of their unpublished chapters. Carla Highkin provided expert assistance in the preparation of parts of this manuscript.

REFERENCES

Azrin, N.H. and Holz, W.C.: Punishment. In Honig, W.K. (ed.): “Operant Behavior: Areas of Research and Application.” New York: Appleton-Century-Crofts, 1966, pp. 380-447.

Barrett, J.E.: Effects of alcohol, chlordiazepoxide, cocaine and pentobarbital on responding maintained under fixed-interval schedules of food or shock presentation. J. Pharmacol. Exp. Ther. 196:

Barrett, J.E.: Behavioral history as a determinant of the effects of d-amphetamine on punished behavior. Science 198:67-69, 1977.

Barrett, J.E.: Nonpharmacological factors determining the behavioral effects of drugs. In Meltzer, H.Y. (ed.): “Psychopharmacology: The Third Generation of Progress. ” New York: Raven Press, 1987,

Barrett, J. E. and Katz, J. L.: Drug effects on behaviors maintained by different events. In Thompson, T. and Dews, P. B. (eds.): “Advances in Behavioral Pharmacology.” Vol. 3. Orlando, Florida: Academic Press, 1981, pp. 119-168.

Barrett, J.E. and Tessel, R.E.: Behavioral pharmacology of drugs affecting norepinephrine transmission. In Zigler, M.G. and Lake, C.R. (eds.): “Norepinephrine: Clinical Aspects.” Baltimore: Williams & Wilkins, 1984, pp. 160-180.

605-615, 1976.

pp. 1493-1501.


Barrett, J. E. and Witkin, J. M.: The role of behavioral history in determing the effects of abused drugs. In Goldberg, S. R. and Stolerman, I. (eds.): “Behavioral Analysis of Drug Dependence.” Or- lando, Florida: Academic Press, 1986, pp. 195-223.

Barrett, J.E., Fleck-Kandath, C. and Mansbach, R.S.: Effects of buspirone differ from those of gepirone and 8-hydroxy-2-(di-n-propylamino)tetralin (8-OH-DPAT) on unpunished responding of pigeons. Pharmacol. Biochem. Behav. 30:723-727, 1988.

Branch, M. N. and Schaal, D. W.: Roles and characteristics of theory in behavioral pharmacology. In Barrett, J. E., Thompson, T. and Dews, P. B. (eds.): “Advances in Behav. Pharmacol.” Vol. 7. Hillsdale, New Jersey: Lawrence Erlbaum Associates, 1990, pp. 171-196.

Branch, M.N., Nicholson, G. and Dworkin, S.I.: Punishment-specific effects of pentobarbital: Depen- dency on the type of punisher. J. Exp. Anal. Behav. 28:285-293, 1977.

Carlton, P.L.: “A Primer of Behavioral Pharmacology: Concepts and Principles in the Behavioral Anal- ysis of Drug Action.” New York: W.H. Freeman, 1983.

Carlton, P.L. : Analysis of physiological mechanism in psychopharmacology . Neuropsychobiology 12:

Cook, L. and Davidson, A.B.: Effects of behaviorally active drugs in a conflict punishment procedure in rats. In Garattini, S. Mussini, E. and Randall, L.O. (eds.): “The Benzodiazepines.” New York: Raven Press, 1973, pp. 327-345.

Dantzer R. : Dissociation between suppressive and facilitating effects of aversive stimuli on behavior by benzodiazepines. A review and reinterpretation. Prog. Neuro-Psychopharmacol. 2:33-40, 1978.

de Carvalho, S.M., de Aguiar, J.C. and Graeff, F.G.: Effect of minor tranquilizers, tryptamine antag- onists and amphetamine on behavior punished by brain stimulation. Pharmacol. Biochem. Behav.

Dews, P.B.: Studies on behavior. I. Differential sensitivity to pentobarbital of pecking performance in

Dews, P.B.: Modification by drugs of performance on simple schedules of positive reinforcement. Ann.

Dews, P.B.: Analysis of effects of psychopharmacological agents in behavioral terms. Fed. Proc. 17:

Dews, P.B.: Drugs in psychology: A commentary on Travis Thompson and Charles R. Schuster’s Behavioral Pharmacology. J. Exp. Anal. Behav. 13:395-406, 1970.

Dews, P.B. and Wenger, G.R.: Rate-dependency of the behavioral effects of amphetamine. In Thomp- son, T. and Dews, P.B. (eds.): “Advances in Behavioral Pharmacology.” Vol. 1. Orlando, Florida: Academic Press, 1977, pp. 167-227.

Geller, I. and Seifter, J.: The effects of meprobamate, barbiturates, &amphetamine, and promazine on experimentally induced conflict in the rat. Psychopharmacologia 1:482-492, 1960.

Graeff, F.G. and de Oliveira, L.: Influence of response topography on the effect of apomorpine and amphetamine on operant behavior of pigeons. Psychopharmacologia 41: 127-132, 1975.

Jeffrey, D.R. and Barrett, J.E.: Effects of chlordiazepoxide on comparable rates of punished and unpunished responding. Psychopharmacology 64:9-11, 1979.

Katz, J. L.: Effects of drugs on stimulus control of behavior. I. Independent assessment of effects on response rates and stimulus control. J. Pharmacol. Exp. Ther. 223:617-623, 1982.

Katz, J. L.: Effects of drugs on stimulus control of behavior. 11. Degree of stimulus control as a determinant of effect. J. Pharmacol. Exp. Ther. 226:756-763, 1983.

Katz, J. L.: Effects of drugs on stimulus control of behavior. 111. Analysis of effects of pentobarbital and &amphetamine. 3. Pharmacol. Exp. Ther. 24676-83, 1988.

Katz, J. L.: Effects of drugs on stimulus control of behavior under schedules of reinforcement. In Thompson, T., Dews, P. B. and Barrett, J. E. (eds.): “Advances in Behavioral Pharmacology,” Vol. 7. Hillsdale, New Jersey: Lawrence Erlbaum Associates, 1990, pp. 13-38.

Katz, J. L.: Drugs as reinforcers: Pharmacological and behavioral factors. In Liebman, J. M. and Cooper, S . J. (eds.): “The Neuropharmacological Basis of Reward.” Oxford: Oxford University Press,

Kelleher, R.T., Fry, W., Deegan and Cook, L.: Effects of meprobamate on operant behavior in rats. J.

Kelleher, R.T. and Morse, W.H.: Escape behavior and punished behavior. Fed. Proc. 235308-817,1964. Kelleher, R.T. and Morse, W.H.: Determinants of the specificity of the behavioral effects of drugs.

158-172, 1984.

15~351-356, 1981.

pigeons depending on the schedule of reward. J. Pharmacol. Exp. Ther. 113:393-401, 1955.

N.Y. Acad. Sci. 65268-281, 1956.

1024-1030, 1958.

1989, pp. 164-213.

Pharmacol. Exp. Ther. 133:271-280.

Ergeh. Physiol. Biol. Chem. Exp. Pharamkol. 6O:l-56, 1968.

408 Witkin and Katz

Laties, V.G.: The role of discriminative stimuli in modulating drug action. Fed. Proc. 34:1880-1888, 1975.

Laties, V. G . and Weiss, B.: Influence of drugs on behavior controlled by internal and external stimuli. J. Pharmacol. Exp. Ther. 152:388-396, 1966.

Leander, J.D.: Effects of punishment attenuating drugs on deprivation-induced drinking: Implications for conflict procedures. Drug Dev. Res. 3:185-191, 1983.

Mansbach, R. S . , Harrod, C., Hoffniann, S. M., Nader, M. A., Lei, Z., Witkin, J. M. and Barrett, J. E.: Behavioral studies with anxiolytic drugs. V. Behavioral and in vivo neurocheniical analyses in pigeons of drugs that increase punished responding. J. Pharmacol. Exp. Ther. 246: 114-120, 1988.

Marr, M. J.: Behavioral pharmacology: Issues of reductionism and causality. In Thompson, T, Dews, P. B. and Barrett, J. E. (eds.): “Advances in Behavioral Pharmacology,”Vol. 7. Hillsdale, New Jersey: Lawrence Erlbaum, 1990 pp. 1-12.

McKearney, J. W.: Effects of d-amphetamine, morphine, and chlorpromazine on responding under fixed- interval schedules of food presentation or electric shock presentation. J. Pharmacol. Exp. Ther.

McKearney, J. W.: Punishment of responding under schedules of stimulus-shock termination: Effects of d-amphetamine and pentobarbital. J. Exp. Anal. Behav. 26:281-287, 1976.

McKearney , J. W. and Barrett, J.E.: Punished behavior: Increases in responding after d-amphetamine. Psychopharmacologia 41:23-26, 1975.

McKearney, J.W. and Barrett, J.E.: Schedule-controlled behavior and the effects of drugs. In Blackman, D.E. and Sanger, D.J. (eds.): “Contemporary Research in Behavioral Pharmacology.” New York: Plenum Press, 1978, pp. 1-68.

McMillan, D.E.: Drugs and punished responding. I. Rate-dependent effects under multiple schedules. J. Exp. Anal. Behav. 19:133-145, 1973.

McMillan, D.E.: Determinants of drug effects on punished responding. Fed. Proc. 34:1870-1879, 1975. Miller, N. E.: Liberalization of basic S-R concepts: Extensions to conflict behavior, motivation and social

learning. In Koch, S . (ed.): “Psychology: A Study of a Science.” New York: McGraw-Hill Book Company, 1959, pp. 196-292.

Morse, W.H., McKearney, J.W. and Kelleher, R.T.: Control of behavior by noxious stimuli. In Iversen, L.L., Iversen, S.D. and Snyder, S.H. (eds.): “Handbook of Psychopharmacology,” Vol. 7. New York: Plenum Press, 1977, pp. 151-180.

Robbins, T. W.: Behavioural determinants of drug action: Rate-dependency revisited. In Cooper, S.J. (ed.): “Theory in Psychopharniacology,” Vol. I . Orlando, Florida: Academic Press, 1981, pp. 1-63.

Sanger, D.J. and Blackman, D.E.: Rate-dependent effects of drugs: Review of the literature. Pharmacol. Biochem. Behav. 4:73-83, 1976.

Schindler, C.W. and Harvey, J. A.: The use of classical conditioning procedures in behavioral pharmacology. Drug Dev. Res., 20:169-187, 1990.

Seiden, L.S., MacPhail, R.C. and Oglesby, M.W.: Catecholamines and drug-behavior interactions. Fed. Proc. 34:1823-1831, 1975.

Sidman, M.: Drug-behavior interactions. Ann. N.Y. Acad. Sci. 65982-302, 1956. Sidman, M.: Some technical problems in evaluating the behavioral effects of drugs. In Votava, Z. (ed.):

Skinner, B. F.: “The Behavior of Organisms.” New York: Appleton-Century-Crofts, 1938. Skinner, B.F.: Animal research in the pharmacotherapy of mental disease. In Cole, J. 0. and Gerard, R.

W. (eds.): “Psychopharmacology, Problems in Evaluation.” Washington, D.C.: National Acad- emy of Sciences-National Research Council, 1959, pp. 224-235.

Skinner, B. F.: “Contingencies of Reinforcement: A Theoretical Analysis. ” New York: Appleton- Century-Crofts, 1969.

Smith, R.F., Gustavson, C.R. and Gregor, G.L.: Incompatability between the pigeon’s unconditioned response to shock and the conditioned key-peck response. J. Exp. Anal. Behav. 18:147-153, 1972.

Teitelbaum, P. and Derks, P.: The effect of amphetamine on forced drinking in the rat. J. Comp. Physiol.

Thompson, D. M.: Stimulus control and drug effects. In Blackman, D. E. and Sanger, D. J. (eds.): “Contemporary Reserach in Behavioral Pharmacology. ” New York Plenum Press, 1978,

190:141-153, 1974.

“Psychopharmacologic Methods.” London: Pergamon Press, 1963, pp. 162-169.

Psychol. 51:801-810, 1958.

pp. 159-238.


Thompson, T.: Behavioral mechanisms of drug dependence. In Thompson, T., Dews, P. B., and Barrett, J . E. (eds.): “Advances in Behavioral Pharmacology,” Vol. 4. Orlando, Florida: Academic Press,

Thompson, T. and Schuster, C.R.: “Behavioral Pharmacology.” Englewood Cliffs, New Jersey: Pren- tice-Hall, 1968.

Thompson, T., Dews, P.B. and McKim, W.A. (eds.) “Advances in Behavioral Pharmacology,” Vol. 3. Orlando, Florida: Academic Press, 1981.

Witkin, J. M. and Barrett, J . E.: Effects of pentobarbital on punished behavior at different shock intensities. Pharmacol. Biochem. Behav. 5535-538, 1976.

Witkin, J . M., Katz, J . L. and Barrett, J. E. Effects of methaqualone on punished and nonpunished behavior. J. Pharmacol. Exp. Ther. 218:l-6, 1981.

Witkin, J. M., Mansbach, R. S., Barrett, J . E. , Bolger, G . T. , Skolnick, P. and Weissman, B. A,: Behavioral studies with anxiolytic drugs IV. Serotonergic involvement in the effects of buspirone on punished behavior of pigeons. J. Pharmacol. Exp. Ther. 243:970-977, 1987.

Witkin, J . M., Lee, M. A. and Walczak, D. D.: Anxiolytic properties of amygdaloid kindling unrelated to benzodiazepine receptors. Psychopharmacology 96:296-301, 1988.

Woods, J.H., Winger G.D. and France, C.P.: Reinforcing and discriminative stimulus effects of cocaine: Analysis of pharmacological mechanisms. In Fisher, S., Raskin, A. and Uhlenhuth, E.H. (eds.): “Cocaine: Clinical and Biobehavioral Aspects.” New York Oxford University Press, 1987, pp.

Wuttke, W. and Kelleher, R.T.: Effects of some benzodiazepines on punished and unpunished behavior

1984, pp. 1-45.

21-65.

in the pigeon. J. Pharmacol. Exp. Ther. 172:397-405, 1970.

Documents

Analysis of behavioral effects of drugs